1. Field of the Invention
The invention relates to the control of internal combustion engines and, more particularly, relates to a method and apparatus for optimizing the excess air ratio (lambda), air charge temperature (ACT), and/or compression temperature of a compression ignition internal combustion engine.
2. Discussion of the Related Art
It is well known that the relative proportion of fuel and air has a marked effect on the combustion process in any internal combustion engine. An engine operating on less than a stoichiometric air/fuel ratio will emit unacceptable levels of unburnt fuel and related emissions. It is for this reason that many engines incorporate measures to supply at least as much air to the engine as is required for stoichiometric combustion. The proportion of air in excess of that required for stoichiometric combustion is known as the excess air ratio or "lambda", which is defined as the ratio of total air available for combustion to that required to burn all of the fuel. It is well known that, if lambda drops below a minimum threshold, oxides of nitrogen (NO.sub.x) and other emissions increase to unacceptable levels.
Current emissions-regulated, gasoline-fueled Otto cycle (spark ignited) engines invariably use full time lambda control. These engines typically use a catalytic converter having a three way catalyst to reduce emissions. In order to permit the three way catalyst to perform in spark ignition engines, lambda is controlled to a value of 1.00 by use of an exhaust oxygen sensor, usually in a closed loop control mode to hold lambda as close to unity (i.e., one or a stoichiometric ratio) as is practical.
It has also been recognized that at least limited lambda control is important in the operation of unthrottled gas-fueled engines. For instance, U.S. Pat. No. 5,553,575 to Beck et al. (the Beck '575 patent) proposes lambda control by skip fire in an unthrottled gas fueled engine with the number of cylinders skipped being calculated to optimize as much as possible lambda under prevailing engine operating conditions. Optimum lambda is calculated experimentally based upon prevailing engine operating parameters including mean effective pressure (MEP), air charge temperature (ACT), intake manifold absolute pressure (MAP), gas fuel charge quantity, ignition timing, exhaust back pressure (EBP), etc. The number of cylinders to be skipped to obtain this lambda then is calculated. That number of cylinders then is skipped in the next combustion cycle. Lambda then is "fine tuned" by varying manifold absolute pressure (MAP). However, skip fire is considered to be the primary mode of control when less than all cylinders are firing.
The Beck '575 patent states that lambda control is considered unnecessary in diesel engines because diesel engines have "an extremely broad range of useful lambdas." The comments in the Beck '575 patent are typical of traditional thinking with respect to diesel engines. For diesel and other compression ignition engines, it is generally assumed that, so long as lambda is high enough, no other adjustment is required. In fact, for compression ignition diesel engines with modern electronic controls, the value of lambda seldom appears in the calibration tables, let alone in a closed loop control strategy. Even those who have recognized some of the benefits of lambda control have failed to recognize the benefits of full time, full range lambda optimization. Hence, while it recently has been recognized that the performance of compression ignition engines can be enhanced by increasing lambda, there is no suggestion in the art to modulate lambda to avoid exceeding an upper limit of lambda.
For instance, SAE Technical Paper 930272 by Hino Motors, Ltd. (the Hino '272 paper) and SAE Technical Paper 931867 by Hino Motors, Ltd. (the Hino '867 paper) recognize that smoke (BSU) emissions and brake specific fuel consumption (BSFC) decrease as lambda increases. Specifically, the Hino '867 paper reported that, as the boost supplied by the turbocharger of a turbocharged diesel engine was increased to increase lambda from 1.6 to 2.2, both BSU and BSFC dropped substantially at a given NO.sub.x emission level. Reduction of BSU with increased lambda and constant NO.sub.x is reflected by the curves 22, 24, 26, and 28 in FIG. 1. Reduction of BSFC with increased lambda and constant NO.sub.x is reflected by the curves 30, 32, 34, and 36 in FIG. 2. The Hino '272 paper reported significant decreases in ignition delay and combustion duration with increased turbocharger boost and consequent increase in lambda. The implicit conclusion reached by both papers was that optimal operation always results from increasing turbocharger boost as much as feasible so as to increase lambda to a maximum practical level. Neither paper recognized that lambda could be too high or that there might be an optimum lambda for a particular engine operating condition that is less than the maximum available lambda, and neither paper sought to modulate a turbocharger or other engine component to optimize lambda on a full time, full-range basis. Nor did either paper discuss the effects of ACT on the operation of a compression ignition engine or the interaction between lambda and ACT.
The effects of lambda variation on a compression ignition engine also were investigated by SAE Technical Paper 870296 to Arnold (the Arnold paper). The Arnold paper discusses the effects of the control of a variable power turbine (VPT) on the performance of a diesel engine. Arnold's experiments began with the mapping of altered boost levels across the engine's speed and load ranges. An array of speed and fuel flows were chosen that covered the lug line from idle to rated speed and also covered loads ranging from 1/4 load to full load from the idle speed to the rated speed. The results of these experiments are summarized in FIG. 3 which illustrates a plot of BSFC against air-fuel ratio at full load. The curves 40, 42, 44, 46, and 48 plot the results at 1750 rpm, 1600 rpm, 1400 rpm, 1200 rpm, and 1020 rpm, respectively. Arnold noted that all of these curves flatten out or reduce slope in roughly the same air-fuel ratio range of 26.5:1 to 31:1.
Arnold concluded that, very much like a gasoline engine, a diesel engine prefers a constant air-fuel ratio and that, while this optimum value varies considerably based on a particular engine design, it usually falls between 26.5:1 and 31:1. Arnold failed to carry his experiments one step further and therefore did not appreciate that deleterious effects occur under some operating conditions if lambda increases above a threshold value. Hence, while the Arnold paper, like the Hino papers, recognized that increasing lambda to something in excess of stoichiometric ratios is desirable during operation of a diesel engine, it failed to recognize that optimum lambda varies with prevailing engine operating parameters including engine speed and that a given air supply system therefore could sometimes supply too much air to the engine under what otherwise might be considered an "optimum" setting. Arnold also failed to address the effects of ACT on engine performance as well as the interplay between ACT and lambda.
Therefore, even in systems such as those disclosed by Hino '867, Hino '272, and Arnold which seek to adjust air supply to enhance engine performance, the air supply typically is adjusted only to be high enough to prevent excessive smoke and BSFC. These and others who have addressed the issue of lambda control failed to recognize that, if lambda rises above a maximum acceptable threshold, incomplete combustion can occur, resulting in excessive unwanted emissions and decreased thermal efficiency. Thus, the search for a truly optimum value of lambda over the entire operating range of the engine has been largely ignored until now. The inventors have recognized that it is essential for optimum control of combustion in an internal combustion engine to maintain lambda values within a permissible range, and preferably to cause lambda values to be adjusted to optimum levels.
ACT control for optimizing engine performance has similarly been ignored or at least underrated. Control of ACT had previously been directed largely to reducing the high temperature emanating from the turbocharger compressor by means of an intercooler. Little attention was given to the possible beneficial effects of decreasing ACT below ambient temperature or of increasing ACT above ambient temperature under certain operating conditions such as light load and/or low ambient temperatures. Conventional diesel engines therefore typically operate at higher than optimum ACT and lower than optimum lambda when at high load and at higher than optimum lambda and lower than optimum ACT when at light load. Consequently, diesel engines have rarely if ever been operated at truly optimum lambda or optimum ACT over the entire engine operating range. In fact, it would be only accidental if the conventional diesel engine were to operate at optimum lambda or optimum ACT values at any operating point in the engine's load/speed ranges.
Some concerted effort will be required to meet future emission regulations for diesel engines, such as EPA 2004 proposed by the United States Environmental Protection Agency. Some of the previously-proposed techniques include 1) exhaust gas recirculation (EGR), 2) particulate traps and, 3) special fuels and fuel additives. All of these techniques are both complex and costly. In addition, all of these techniques are directed more at correcting the deficiency (inadequate lambda control) rather than preventing the deficiency from occurring in the first place. It is not yet appreciated that a combination of full time, full range lambda control, improved fuel injection, and improved combustion temperature control through ACT control has the potential to obviate the need for these additional corrective techniques. Even if some of these corrective techniques are used, it appears logical that the optimization of lambda and ACT should be accomplished prior to the addition of some of these more severe techniques.